>       msvcr80.dll!__crt_debugger_hook(int _Reserved=)  Line 65        C
>       msvcr80.dll!_invalid_parameter(const wchar_t * 
> pszExpression=0x00000000aff9055c, const wchar_t * 
> pszFunction=0x0000000000000000, const wchar_t * pszFile=0x0000000000000000, 
> unsigned int nLine=6, unsigned __int64 pReserved=0)  Line 212      C++
>       msvcr80.dll!signal(int signum=1245184, void (int)* 
> sigact=0x0000000000130000)  Line 419 + 0x23 bytes    C
>       libeay32.dll!pushsig()  Line 606 + 0x11 bytes   C

 > Theoretically, it could be problem with VC8 compilers on Windows-64A,
 > although not very likely.

Depends on defintion of "problem." When Microsoft decides to throw an 
fatal exception as above on unimplemented signal number passed to signal 
function, instead of just returning an error code like they and the rest 
of the world used to for many years. Is it a problem? Because that's 
what happens. But unlike notorious str* thing, when it's possible to 
regain compatibility by defining _CRT_SECURE_NO_DEPRECATE, it doesn't 
seem to be possible to revert to original signal behavious, as exception 
is hard-coded into msvcr80.dll. They say one can install 
_set_invalid_parameter_handler, but it's hardly good style for library 
to steal handlers application might have set and it's not clear how to 
detect presense of new run-time [compiler version alone is not 
sufficient, as for example Platform SDK comes with new compiler and 
elder MSVCRT].

http://cvs.openssl.org/chngview?cn=14451 and case is being dismissed. A.

______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
Development Mailing List                       [email protected]
Automated List Manager                           [EMAIL PROTECTED]

Reply via email to